Skip to content

module adabmDCA.api.diagnostics

Turn the files of a training run (and optionally a sampling check) into a verdict.

The rules follow the "Diagnose a run" guide (docs/algorithms/diagnostics.md): each finding names the failure it points to (F1-F8) and the change the guide recommends, written as CLI options. Thresholds are the guide's; they are heuristics, so findings say what was measured.

Global Variables

  • GUIDE
  • FAILURES
  • SEVERITY_ORDER
  • STALE_SECONDS
  • MIN_UPDATES_FOR_RATES

function diagnose_run

diagnose_run(
    run_dir: 'str | Path',
    label: 'str | None' = None,
    sampling_json: 'str | Path | None' = None
) → dict[str, Any]

Indicators and findings for a training run directory, and optionally a sampling check.

This file was automatically generated via lazydocs.